Neo4j configuration

Neo4j is the Perforce IPLM database backend. Find the Neo4j configuration options.

Perforce IPLM-specific settings can be defined in the neo4j.conf file. These settings should be identical for all the Neo4j servers in a cluster. The Neo4j server must be restarted for a change in the settings to take effect. If an illegal value is specified for a setting, the Neo4j server startup will be stopped and an error will be logged in the Neo4j log file.

Perforce IPLM settings

Setting Type Description Default value
mdx.repo_path_validation_enabled boolean

If true, repo paths must not overlap across all the IPs

Example: /workspaces/abc and /workspaces/abc/xyz cannot both exist

 true
mdx.hook_script string

Absolute path to the server hook script

The server hook script will be executed before some operations. It is useful to implement custom validation rules.

Example: /methodicsiplm/bin/server-hook.sh

 n/a
mdx.hook_execution_timeout integer

Server hook execution timeout (in ms, must be >= 0)

If executing the server hook script takes longer than this time, it is aborted and the operation fails.

 5000
mdx.hook_lock_wait_timeout integer

Server hook lock wait timeout (in ms, must be >= 0)

At most one instance of the server hook script is executed for any given object (Library, IP, etc.).
If another operation on the same object needs to execute the server hook script, it will wait for the first one to be over.
If the first execution doesn't finish in the specified time, the second operation fails.

 3000
mdx.pagination_cli_page_size integer

Page size used by the CLI (0 means unlimited, must be >= 0)

It is recommended to set this setting to a value greater than 1000.

 5000
mdx.pagination_cli_concurrent_requests integer

Number of concurrent requests for pages made by the CLI (must be > 0)

It is recommended to set this setting to the number of slave servers in the Neo4j cluster.

 5
mdx.user_session_last_access_timestamp_tolerance integer

Tolerance period between updates of the user session last access timestamp (in ms, must be >= 0)

In order to minimize lock contention and improve performance, the user session last access timestamp is updated only if it is older than this tolerance period.
That timestamp is used primarily to decide if the license can be detached from the user and attached to another user.

 48 hours
mdx.transaction_max_retries_entity_not_found integer

Maximum number of transaction retries when an entity has been deleted in a concurrent transaction (must be >= 0)

When an object is deleted and listed simultaneously, the listing transaction can be retried.
This setting specifies the maximum number of retries.

 3
mdx.transaction_max_retries_other integer

Maximum number of transaction retries when another error occurs (must be >= 0)

When a transaction fails because of an unexpected error, it can be retried.
This setting specifies the maximum number of retries.
It is recommended to leave this setting to 0.

 0
mdx.transaction_retries_delay integer

Delay before each transaction retry (in ms, must be >= 0)

Setting this setting to a value greater than 0 increases the chance to successfully execute the transaction, at the cost of a possibly longer response time (due to the delay).

 0
mdx.janitor_initial_delay integer

Delay before the first execution of the janitor (in seconds, must be >= 0)

This delay helps prevent the janitor from running during the initial startup of Neo4j.

 60
mdx.janitor_period integer Time between executions of the janitor (in seconds, must be > 0) 24h
mdx.file_list_compressor_threads integer

Number of threads used for background file list compression (must be >= 0)

It is recommended to set this setting to the number of cores of the server.

 1
mdx.file_list_compressor_batch_size integer

Maximum number of uncompressed file lists scheduled for compression during one janitor execution (must be > 0)

The janitor compresses file lists in batches.

 50000

Useful Neo4j settings

Setting Type Description Default value
dbms.threads.worker_count integer

Number of Neo4j worker threads. Defines the size of the pool thread available to the database engine to execute queries. By default it is set to the number of cores the OS reports (so typically, 4-8 for low-end machines, 16-32 for beefier ones); up to 500, with the maximum custom value can be 40K+ . So that number is really a function of the CPU profile and workload.

If the workload causes the DB engine to trigger queries that block; it makes sense to bump up the thread count so that work  can be done by other threads. If the workload generates expensive queries that just take a long time to run and if your CPU is maxed out, then there is no point in bumping this up anymore and you just need to add another machine to the cluster or get a more powerful machine with more cores. Since those are threads, the memory accounting will happen towards the main java process, which should be auto-tuning; based on available memory in the system.

 number of cores

Types of values

Type Description
boolean

One of:

  • true, yes, on, 1
  • false, no, off, 0
integer An integer number in decimal format
string A string